i tend to get incomplete responses when i use the mythomax and airoboros models on agnai.chat - short outputs are fine but for some reason it always cuts off at around 200 words? i don't think it's an issue with my context settings because i tried shortening it well below the maximum, but the problem persists.
#Responses cutting off
arctic parrot
knotty arch
Same problem with almost any of the open source or even anthropic, I think there is a max token limit or something in the API call, because it works perfectly fine when I try on the playground with larger max tken limits
pastel walrus
Mythomax and Airoboros have 300 max tokens per response; those are the model's limitations.